Computing devices may perform certain functions in response to audio commands indicated by user speech. To provide this functionality, a computing device may include or be operatively coupled to an audio input device such as a microphone. In some known scenarios, a computing device may analyze audio input received from a user (e.g., via a microphone) to infer an audio command indicated by the audio input and, in response, interrupt the user by outputting an audio indication of the audio command. Such an interruption can be distracting or frustrating to the user, however. In alternative scenarios, the computing device may maintain the microphone in a constant, active listening state until an audio command is received, but this feature may impose a significant drain on the computing device's battery. Further, as the set of possible audio commands interpretable by the computing device can be difficult for users to discover, users may be forced to rely on memory when attempting to utilize such audio command functionality, thus making such functionality difficult to use and of limited utility to the user.